What is the added value of Bone Mineral Density as a risk factor to Fracture Risk?

Talk Code: 
Paula Dhiman
Nadeem Qureshi, Stig Andersen, Tahir Masud
Author institutions: 
University of Nottingham, Aalborg University Hospital


Fragility fractures are a major and increasing cause of morbidity; approx. 300,000 of such fractures are presented in UK hospitals, per year. With an expected rise in incidence parallel to the world’s ageing population, the potential cost resulting from these fractures is also projected to rise to £2.2 billion in 2025. Thus, prevention is a key priority. The National Institute of Clinical Health and Excellence (NICE) recommend using the QFracture® and FRAX® risk prediction tools to assess patient risk of a fragility fracture, and offering preventable treatment. Although, osteoporosis (bone mineral density (BMD) measurement, T- score≤-2.5) has shown to lead to an increased susceptibility of fragility fractures, and is associated to therapeutic options such as bisphosphonates, BMD is treated as an optional fracture risk factor. However, NICE recognise the gap in the evidence and have recommended research assessing the added value of BMD as a risk factor in fracture risk assessment.


A prospective cohort study was conducted using patients from the Aalborg University Hospital Record for Osteoporosis Risk Assessment (AURORA) dataset; patients were followed up from 01/01/2012 to 01/01/2014, for an osteoporotic fracture using the National Patient Registry of Denmark.Data was split to derive and validate the fracture risk tools. Cox regression analysis was used to derive the fracture risk prediction tools, with and without BMD. The performance of each model was assessed using measures of calibration, discrimination, and reclassification.


A significant association of bone mineral density was found with the fracture outcome (p<0.001). Adjusting for other risk factors, osteoporotic patients had 75% higher hazard of a fracture (HR: 1.75 (95% CI: 1.28 to 2.38)) compared to osteopenic/normal patients; and 40% lower hazard per unit increase in T-score (HR: 0.60 (95% CI: 0.52 to 0.69)).The validation study showed a 1.15% increase in Harrell’s C-Index (0.7640 to 0.7728) when adding continuous BMD to fracture risk prediction; but a (0.62% decrease when adding binary BMD. Reclassification tables showed adding binary BMD shifted 109 patients into different risk categories; 65% of these were reclassified correctly. Adding continuous BMD shifted 463 patients into different risk categories; 60% of these were reclassified correctly.


Bone mineral density is beneficial to fracture risk prediction and should be considered in routine fracture risk assessment. Performance measures showed using BMD in a continuous format is better than using a binary format; categorising t-score values according to current guidance. The added value of BMD to fracture risk prediction should be investigated using routinely collected primary care data in other countries.

Submitted by: 
Paula Dhiman
Funding acknowledgement: 
NIHR - SPCR Nottingham Hospitals Charity